Home / Research / Federal AI Research Grant Statistics

Market research · October 22, 2026

Federal AI Research Grant Statistics: 2026 Report

By Darin ten Bruggencate, Founder, AI Industry Reviews

The National Science Foundation (NSF) and the National Institutes of Health (NIH) are two of the largest public funders of US university and lab research. When they put money into projects that use artificial intelligence or machine learning, that funding shapes what academic labs can build and what trainees learn. In 2025 and 2026, a separate story ran alongside new awards: public trackers documented grants that were terminated, frozen or later restored.

We wanted a clear count of AI research awards at NSF and NIH for FY2024 through FY2026, and of AI grants that Grant Witness recorded as disrupted.

Our analysts pulled three public sources on October 19, 2026. From NIH RePORTER, we took awards whose project titles contain AI or machine-learning phrases. From USAspending.gov, we took obligations on NSF awards whose records match AI keywords. From Grant Witness, a public tracker of terminated and frozen federal research grants, we took NSF and NIH grants whose titles, abstracts or topics match AI phrases. NIH award dollars for AI-titled projects fell 23.2% from FY2024 to FY2026, while NSF obligations on AI-keyword awards rose in each of those years.

In the following sections, we report NIH AI awards by fiscal year, NSF AI-keyword obligations, the Grant Witness disruption and restoration counts, and the NSF directorates where terminated AI grants clustered.

NIH AI Research Awards by Fiscal Year

Each NIH RePORTER row is a fiscal-year award record. We kept awards whose project title contains one of the AI phrases listed in Methodology.

NIH AI Research Awards (Title Phrase Match), FY2024-FY2026

• NIH title-matched AI award dollars fell in both years, by $48.4 million in FY2025 and by another $51.1 million in FY2026.
• Award counts stayed in a narrow band: 725 in FY2024, 734 in FY2025 and 684 in FY2026.
• Because counts stayed in that band while the average award fell 18.6%, from about $591,000 to about $481,000, most of the drop in dollars came from smaller awards.

NSF AI-Keyword Grant Obligations by Fiscal Year

NSF does not publish a single “AI award” flag. We used USAspending.gov obligations for NSF assistance awards whose records match the keywords “artificial intelligence” or “machine learning,” summed by federal fiscal year. This keyword rule is broader than a title-only rule, so the dollars are larger than a title-only NSF count would be.

NSF Obligations on AI-Keyword Assistance Awards, FY2024-FY2026

• NSF obligations on AI-keyword awards rose from $1.51 billion in FY2024 to $2.02 billion in FY2026.
• Most of the increase came in FY2025, when obligations rose 25.6%. Growth slowed to 6.2% in FY2026.
• The NSF and NIH series are defined differently, so their dollar figures should not be added together as one federal AI research budget.

AI Grants Terminated or Restored in Grant Witness

Grant Witness compiles disrupted federal research grants from agency files, USAspending, court records and other public sources. We filtered its NSF and NIH tables to grants whose titles, abstracts or topics match AI phrases, then split them by current status.

AI-Matched Grants in Grant Witness, as Retrieved October 19, 2026

• All 111 NSF AI-matched grants that remain disrupted carry a detailed status of Terminated in the Grant Witness file. Their funds promised sum to $81.3 million.
• 142 NSF AI-matched grants were restored (127 confirmed reinstatements and 15 unconfirmed), with $111.2 million in funds promised.
• 86.8% of the NIH AI-matched grants in the file are restored (334 of 385), compared with 56.1% at NSF (142 of 253).
• Grant Witness statuses change as agencies reinstate awards, so some of the 162 AI-matched grants still listed as disrupted (111 at NSF and 51 at NIH) may move to restored in later updates.

Where NSF AI Terminations Clustered by Directorate

Among the 111 NSF AI-matched grants that remain disrupted with terminated status, STEM Education, Social, Behavioral and Economic Sciences, and Computer and Information Science and Engineering account for most of the cases.

NSF AI-Matched Terminated Grants Still Disrupted, by Directorate

• STEM Education accounts for 29 of the 111 still-disrupted NSF AI-matched terminations, or 26.1%.
• Computer and Information Science and Engineering accounts for 22, or 19.8%, fewer than STEM Education or Social, Behavioral and Economic Sciences.
• Together, STEM Education and Social, Behavioral and Economic Sciences account for 52 of the 111 (46.8%), close to half of the still-disrupted NSF AI terminations.

What Federal AI Research Funding Trends Mean Going Forward

For university and hospital labs that do AI research, new federal money has grown at NSF and shrunk at NIH since FY2024. A lab whose AI work fits NSF programs has had more NSF money to compete for, while biomedical AI projects at NIH have received smaller awards on average.

Award totals do not show what happened to grants after they were made. Grant Witness lists 638 AI-matched NSF and NIH grants that were terminated or frozen in 2025 and 2026. Most were later restored, and 162 were still disrupted when we retrieved the files.

So if you plan university AI research budgets or partnerships, separate new-award totals from disruption status, and check Grant Witness for any active NSF or NIH AI grant you depend on.